Vallard Benincosa

Being in your 50s is honestly hilarious. 😂 One friend is a grandma, another still has a 10-year-old at home. One is celebrating 25 years of marriage, while another is dating someone who’s 25. One looks like they haven’t aged since 35, and another threw their back out just by sneezing. Some are planning retirement, others are starting over after divorce. Some are becoming empty nesters, while others are still attending school meetings and making lunches. Some are booking cruises, others are discovering dating apps for the first time. Half of us are saying, “Life is just getting started!” while the other half need three business days to recover from sleeping in the wrong position. We’re all the same age, yet somehow living completely different lives. 😂 Nobody warned us that our 50s would look this chaotic, unpredictable, and entertaining. Honestly, we deserve a group discount on therapy, multivitamins, and heating pads. 😭

Today, we’re open sourcing the full NVIDIA Run:ai platform as part of the NVIDIA DSX AI Factory Platform. Developers, platform teams, and partners can build on production-grade AI workloads and GPU orchestration to improve GPU utilization, increase workload throughput, and bring AI factory solutions to market faster. Deploy the integrated NVIDIA Run:ai platform or adopt individual components to fit your needs. See what’s included, how to get started, and how to contribute: nvda.ws/4hRTubt

This is notable. DeepSeek, a lab usually first to pioneer novel algorithms and architectures, is saying that at this point, the ROI of improving data quality far exceeds that of working on novel post-training algorithms. I think this has already been true for some time for non-lab practitioners. If you're doing llm post-training, 80% of your effort should go into looking at your data. This means: - Hiring experts to dig through your RL tasks - Sifting through rollouts and sft data by hand to remove suspicious samples. Make sure all tasks are actually passable. - Making sure your data is diverse in both difficulty and category.

I resigned from Anthropic today. I spent the last three years doing pretraining research at both OpenAI and Anthropic. Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives. More thoughts below.

"We are experiencing a higher than normal call volume..." I don't think they know what "normal" means. If every moment of the day is "higher than normal" then that is, in fact, normal.

have you ever noticed that beautiful people give compliments, happy people celebrate the success of others, multilingual people don’t mock someone else’s accent, educated people share their knowledge, and confident people don’t compete... they inspire!

notice how i didn’t interrupt the ladies sunbathing and simply minded my own business? take notes, fellas.

That’s it, for the rest of the Cup I’m watching everything on Telemundo. This destroys the Fox crew who successfully put me to sleep before this even happened tonight.

Every company’s AI workflow rn be like 😭💀

🗿🗿

A vision of a modern Tenochtitlan in 2026, in a timeline where the Aztec Empire repelled Spanish conquest and modernized on its own terms. Enjoy: